redth/engraim/engraim-calibrate
Claude Code and agent skill by redth/engraim
- community
Checking that the SKILL.md file is still in the repository.
Record and manage how Claude should BEHAVE in this workspace — corrections, conventions, do/don't rules, tone. Use WHENEVER the user corrects you, states a preference, expresses frustration with an approach, or a `verdict:` annotation comes back negative. Also use to reaffirm, promote, or retire existing behavior overrides. This is the workspace's self-improving frame; it's distinct from facts about the world (use engraim-memory for those).
Install
npx vskill@latest install redth/engraim/engraim-calibrateSKILL.md
Calibrate: the workspace's behavioral frame
Corrections about how to act here are first-class memory. Capturing them is what makes the workspace get easier to work in over time.
When to calibrate
- The user corrects your approach ("don't do X", "always Y here").
- They state a standing preference (tooling, style, process, tone).
- They push back or express frustration — the underlying preference is the override.
- A
verdict:you left on prior work comes back negative — turn the lesson into a rule.
A UserPromptSubmit hook already watches for durable, project-wide phrasing ("from now on", "always remember", "we should do it like this", "in this project …") and injects a [EngrAIm] nudge when it fires. Treat that nudge as a prompt to act here — but apply your own judgment: capture it only if it's a standing rule for the workspace, not a one-off for the current task. The hook detects; you decide.
How
- Phrase it as a short imperative rule: "run tests before committing", "prefer pnpm",
"keep answers terse, no preamble". Add a scope tag (git, testing, tone, …) when domain-specific.
- Call
calibrate(rule, scope?, reason?). Re-stating a rule you already have REAFFIRMS it
(refreshes its 90-day expiry and counts toward promotion) — that's good, not a duplicate.
The promotion ladder
- Logged / one-off — captured, low commitment.
- Standing (default) — active and injected into every session; expires in ~90 days unless
reaffirmed. This is the normal home for a preference.
- Permanent — graduated via
promote_override(id); no expiry, written intoframe/SOUL.md.
Do this deliberately, when a preference has proven durable (often promotion_candidate: true). Workspace rules are written directly; a change to a shipped plugin skill is proposed as a PR instead, never auto-applied.
Discipline
- Expiry is a feature. If a rule no longer applies, let it lapse — don't reaffirm it. Use
retire_override(id) to drop one immediately.
- Reaffirm honestly. Only reaffirm when the rule actually held up again, so the promotion
signal stays meaningful.
- Don't pre-empt the user. Record what they actually asked for, not an inflated version.
Security
- Security scanPassed, 90 of 100Pattern checks on the skill's files.
- Model reviewPassed
Quality evaluation
- Task evaluationNot runNo evaluation yet. Run one in Skill Studio.
Source
- SKILL.md in the repositoryCheckingplugins/engraim/skills/engraim-calibrate/SKILL.md
Popularity
- Repository stars
- 1
- 7-day trend score
- 0
Stars and forks are for the whole GitHub repository, not this skill alone. Refreshed 21 Oct 2026, 14 days ago.
Works with
Any agent vskill installs to: 56 targets in all. These 39 have a page on this site.
- Claude Code
- Cursor
- GitHub Copilot
- Windsurf
- Codex
- Gemini CLI
- Cline
- Amp
- Roo Code
- Goose
- Continue
- Kilo Code
- Junie
- Trae
- OpenHands
- Qwen Code
- AI Agent
- Tabnine Chat
- Sourcegraph Cody
- Amazon Q
- Firebase Genkit
- Sweep
- Devin
- Phind
- Mentat
- Claude Desktop
- Aider
- GPT Pilot
- Plandex
- bolt.new
- v0
- Replit Agent
- Anse
- TypingMind
- BetterChatGPT
- LibreChat
- Warp AI
- Pieces
- Fig AI
Embed badge
[](https://verified-skill.com/skills/redth/engraim/engraim-calibrate)